This is the thread for providing feedback on the IPFS SourceCred demo, available at https://decentralion.github.io/sourcecred-demos/ipfs/prototype/.
Great work so far! Some feedback:
I think that comments need a better way to be valued. Emoji responses will likely help here, but also I think that increasing a comments weight if its author has someone respond to them later in the thread.
Also pull requests should be weighted better, I can’t quite pinpoint what the feature is, but weighting big bugfixes and heavily requested features quite a bit higher than little docs fixes and smaller bugfixes. (not that doc fixes and small bugs arent valuable, i just think that the difference in value isnt high enough in the current algorithm)
Thanks for the suggestion! That should be pretty straightforward to implement. I’ll put it in shortly and we can see how it affects the cred: #804.
So the question is: how can we distinguish those pulls? Some ideas we’ve considered:
- Use labels in the GitHub data model #483 (create edges to the label itself? or as a weight modifier?)
- See which files the pull modified (create edges to the files it modified—and the file is connected to the directory—and certain directories may have specified weights)
- See how many lines of code the pull modified
Curious if any of those sound particularly promising to you.
Feels like having PRs that close many issues by being merged are more valuable than PRs with no association to other issues or a smaller amount and can be used to see the value of a PR and link it to other resources.
- See how many lines of code the pull modified
This feels dangerous to include in the calculation, unless it’s less lines modified = higher value
which isn’t correct either I think. Smaller changes with issues closed because of merge I think are more valuable than big changes that closes issues. But also, removals in a PR is more valuable than modifications or additions.
^ just a bunch of quick thoughts